top of page

Voiceover Too Quiet? Diagnose the Cause Before Raising the Volume

Oct 20
4 min read

Updated: Nov 5

If a voiceover is too quiet, first listen to it alone and compare it with a known working file on the same device. The cause may be playback volume, a low track level, music masking the words or uneven speech dynamics. Each needs a different adjustment. Raising everything together will not make narration clearer over music.


Rule out a playback problem


Check device volume, player mute, the selected audio output and any application-level volume control. Try headphones and an ordinary speaker. If other familiar files are also quiet, investigate the listening setup before changing the narration itself.


Open the original downloaded voiceover outside the video editor. If it sounds normal there but quiet in the project, inspect clip gain, track faders, automation and routing. A voice track accidentally reduced by an automation point can look intact while playing much too softly.


Keep the original unchanged while testing. This gives you a reliable comparison if several adjustments make the edit confusing.


Listen to the voice without music


Mute the background tracks and listen through the narration. If the speech is clear alone but disappears when music returns, the mix is the primary issue. Lower or reshape the competing background rather than immediately adding gain to an already strong voice track.


Music with a busy vocal or strong midrange can compete with speech even when it does not seem very loud. Try a less crowded section or a different arrangement, and check the words at a normal listening level. The audience should not need to concentrate to recover a sentence.


For example, suppose a narration peaks close to the available ceiling but the soundtrack still covers consonants. Raising the voice may cause clipping without solving the competition. Reducing the music during speech and checking the mix on a phone speaker is a more targeted experiment.


Understand gain and normalization


Gain raises or lowers the signal level. If the entire voice is consistently low and has adequate headroom, a gain increase may be sufficient. Watch the level meter and listen for distortion; do not keep raising a track that is already reaching the available peak ceiling.


Peak normalization adjusts a file according to its highest peak. One isolated loud syllable can limit how much the rest rises. Loudness normalization targets a measure of perceived program loudness instead. These are different tools and can produce different results on the same narration.


Do not apply a universal loudness number to every project. A platform, broadcast client or phone system may specify its own delivery standard. Follow that specification when one exists, and check both perceived loudness and peaks after processing.


Why is speech quiet when the rest of the video is loud?

Solo the voice, then replay the same line in the mix. If it is clear in isolation, lower the music or effects under the line before raising the entire output. If it is quiet even in isolation, check the voice track gain first.

Check speech in the busiest part of the mix

Play the quietest sentence with the music and effects active. If the dry voice is clear but gets buried in the mix, lower the competing tracks first and check that the final export still meets its delivery limits.

Use compression for uneven delivery, not as a blind boost


If some phrases are much louder than others, compression can reduce the difference between strong and quiet passages. It acts according to settings such as threshold and ratio. It may reduce level, so makeup gain or a later normalization step can be necessary.


Start gently and compare with the original at a similar listening level. Listen for pumping, exaggerated breaths and a loss of natural emphasis. A quieter sentence may need a small clip-level adjustment rather than heavy compression across the whole track.


For a spoken explanation, preserving clear words and comfortable dynamics matters more than making the waveform look uniformly solid. If a word is missing, badly generated or already distorted, level processing cannot reconstruct the intended performance reliably.


Treat limiting as a final guard, not a repair


A limiter constrains peaks more strongly and can help control the final output when used appropriately. It does not repair audio that was already clipped in the source. If the original has harsh flattened peaks or audible distortion, return to a clean source or replace the affected passage when possible.


Do not normalize, compress and limit repeatedly without a reason. Keep a simple processing chain, compare stages and stop once the delivery meets the specification and sounds clear. More processors can create more problems, especially when each is reacting to changes made by the previous one.


Check the actual export


Export the intended delivery and play it outside the editor at normal speed. Listen with and without headphones, then on a typical small speaker. Check the first word, quiet phrases, loud transitions and the final sentence. If narration is part of a video, inspect the destination’s processed upload as well.


Use the same listening volume for comparisons. A louder file can seem more impressive initially, even if it is harsh or tiring. The useful test is whether the audience can understand the message comfortably across the complete piece.


AI LEMNISCATA’s voiceover workflow currently delivers MP3 narration. Mixing, loudness adjustment and video assembly are separate editing tasks. Do not assume that downloading a new file will automatically match a particular platform’s final mix specification.


Common questions


Why is my voiceover still quiet after peak normalization?


An isolated peak may already be high while the rest of the speech remains comparatively low. Inspect uneven dynamics and consider a measured combination of clip adjustments, compression or loudness normalization. Check the destination’s requirements before choosing a target.


Should I turn the music down or the voice up?


Listen to the voice alone first. If it already has a healthy level and sounds clear, reduce the competition. If the voice itself is consistently low with safe headroom, adjust it too. Judge the finished mix, not either fader position in isolation.


Can I fix clipped audio by lowering the volume?


Lowering the level makes existing distortion quieter; it does not undo the damage. Look for an unclipped original or replace the affected audio. Avoid building the final mix from a damaged intermediate file.


If raising the level produces overload or the original already sounds damaged, follow Clipped or Distorted Voiceover? Find Where the Damage Begins.


Related guides and references



Audacity: loudness normalization distinguishes loudness and peak approaches. Audacity: reducing dynamic range explains compression and limiting.


bottom of page